List of Abbreviations

(1)

DIABETES EDUCATION CENTRE ATTENDANCE AND THE EFFECT ON MEDICATION UTILIZATION IN THE ELDERLY IN ONTARIO

in partial fulfillment of the requirements for the degree of

Master of Science, Medicine (Clinical Epidemiology) Faculty of Medicine

Memorial University of Newfoundland

October 2015

St. John’s Newfoundland

(2)

ii

ABSTRACT

Diabetes education centres (DECs) provide patients with self-management skills to control diabetes and manage complications. To evaluate the effect of DEC attendance on prescriptions for diabetes treatments, prescriptions for cardiovascular risk reduction, and visits for retinopathy screening, a population based cohort study of residents of Ontario, Canada with diagnosed diabetes aged ≥65 years was performed using

administrative databases. DEC attendance was identified using a registry of visits to all DECs in the province in 2006. Demographic and clinical confounders and pre-index utilization were used to adjust the logistic regression and also to construct a propensity score matched cohort. Patients attending DECs had greater filling of prescriptions for statins than non-attendees in both analyses. DEC attendance was also associated with greater drug dispensation of glucose lowering medications, glucose monitoring strips and ACE inhibitors/ARBs, and visits to ophthalmology/optometry in both analyses. Diabetes self-management education at DECs is associated with better quality of care in the elderly in Ontario.

(3)

iii

ACKNOWLEDGEMENTS

There are many people I would like to acknowledge in helping me complete this thesis. Dr. Baiju Shah, my supervisor at the University of Toronto and the Institute of Clinical Evaluative Sciences (ICES), for his endless patience and support during my project and over the years it took for this thesis to finally be completed; Dr. Sean Murphy, my co-supervisor at Memorial and committee member Dr. Bryan Curtis at Memorial for their advice in the completion of this thesis; the Internal Medicine Residency program at Memorial for allowing me time to complete the course work required for this MSc; the Endocrinology Fellowship program at the University of Toronto for the time and support needed to complete my research project. I would also like to thank the staff at ICES especially Aamar Manzoor for assistance with statistical analysis of data and Karen Cauch-Dudek, project manager for the Diabetes Education Centre registry.

(4)

iv

List of Figures and Tables

Figure 2.1: Study timeline ... 40 Table 3.1: Baseline characteristics at index date ... 45 Table 3.2: Association of medication dispensation and physician visits with DEC

attendance by logistic regression ... 48 Table 3.3: Baseline characteristics of cohort post propensity score match ... 50 Table 3.4: Absolute rates of medication and healthcare utilization in the propensity score matched cohort after the index date ... 52

(7)

vii

List of Abbreviations

A1C glycated hemoglobin

ABCD Appropriate Blood Pressure Control in Diabetes

ACCORD Action to Control Cardiovascular Risk in Diabetes Study ACE angiotensin converting enzyme

ADVANCE Action in Diabetes and Vascular Disease: Preterax and Diamicron Modified Release Controlled Evaluation

ARB angiotensin II receptor blocker

ASCOT-LLA Anglo-Scandinavian Cardiac Outcomes Trial – Lipid Lowering Arm BMI body mass index

CARDS Collaborative Atorvastatin Diabetes Study CI confidence interval

CIHI Canadian Institute for Health Information DCCT Diabetes Control and Complications Trial DEC Diabetes Education Centre

DESMOND Diabetes Education and Self-management for Ongoing and Newly Diagnosed type 2

d.f. degrees of freedom

DIN Drug Identification Number

DSME diabetes self-management education

EDIC Epidemiology of Diabetes Interventions and Complications study

(8)

viii HDL high density lipoprotein

HOT Hypertension Optimal Treatment HPS Heart Protection Study

HR hazard ratio

kg kilogram

LDL low density lipoprotein

LHIN Local Health Integration Network MI myocardial infraction

mm Hg millimeters of mercury mmol/L millimole/litre

NHANES National Health and Nutrition Examination Survey NPHS National Population Health Survey

ODB Ontario drug benefits program database ODD Ontario Diabetes Database

OHIP Ontario Health Insurance Plan

OR odds ratio

QALY quality adjusted life year RCT Randomized control trial RPDB Registered Persons Database SMBG self-monitoring of blood glucose TNT Treating to New Targets

UKPDS United Kingdom Prospective Diabetes Study VADT Veteran’s Affairs Diabetes Trial

(9)

1

Chapter 1: Introduction

1.1 Diabetes Burden and Prevalence

Diabetes mellitus is a metabolic disease of hyperglycemia from defective insulin secretion, or action or both (Goldenberg et al., 2013). The long-term effects of this hyperglycemia are associated with the damage, dysfunction and failure of many organs including the kidneys, eyes, nerves, heart and blood vessels. Diabetes is classified into various types, the most common of which are type 1 and type 2. Type 1 diabetes is a result of pancreatic beta cell destruction from autoimmune or idiopathic process, leading to a lack of insulin secretion and susceptibility to ketoacidosis. Type 2 diabetes is mainly a problem of insulin resistance with relative insulin deficiency. Type 2 diabetes is more common than type 1 diabetes and comprises the majority of the diabetes population greater than 65 years of age.

The prevalence of diabetes is increasing in Ontario, with the age-adjusted and sex- adjusted diabetes prevalence increasing by 69%, from 5.2% in 1995 to 8.8% in 2005 (Lipscombe et al., 2007). The prevalence of diabetes in adults greater than 50 years of age increased from 10.6% in 1995 to 17.1% in 2005, a prevalence rate increase of 62.8%. The prevalence of diabetes is 20% in women and 25% in men in the over 65 population in Ontario (Creatore et al., 2010). The cost of diabetes in Ontario is estimated at $4.9 billion in 2010 and expected to increase to over $6.9 billion by 2020 (Canadian Diabetes

Association, 2009). As the prevalence and cost of diabetes increases in Ontario, the need to provide quality care through interventions such as diabetes self-management education (DSME) also increases.

(10)

2 1.2 Diabetes Education

Management of diabetes, as in many chronic diseases such as asthma and rheumatoid arthritis, is dependent on the responsibility patients take in their own care (Newman et al., 2004). Patient involvement in the management of their care is termed self-management and includes the patients’ ability to manage the symptoms, treatment, physical and psychosocial effects and lifestyle changes essential to living with a chronic condition. For self-management to be effective it must include not only the ability to monitor ones condition and follow the treatment guidelines but to institute the

psychological and social changes of living with a chronic illness to manage the effect on their lives.

The chronic disease self-management program, a community based patient self- management education course involves three principal assumptions: different chronic diseases have similar self-management problems and disease-related responsibilities;

patients can take responsibility for the daily management of their disease; and patients practicing self-management will have improved health status (Lorig et al., 1999). This model has been shown to increase healthful behaviors and maintain or improve health status and decrease rates of hospitalization in a heterogeneous group of diseases.

Diabetes education provides patients with self-management skills necessary for management of diabetes such as diet and lifestyle changes, medication compliance, and self-monitoring of blood glucose (SMBG) (Ismail et al., 2004). The self-care

responsibilities for optimum control include modification of lifestyle with diet, exercise, and weight loss, SMBG, foot care, and the administration of oral medications and insulin injections. The objectives of DSME are to increase individual’s involvement, confidence

(11)

3 and motivation for control of their diabetes. Diabetes education for self-management is a fundamental component of diabetes care and most beneficial when working in

conjunction with the healthcare team (Jones et al., 2013). It can be individualized to patient’s metabolic stability, treatment recommendations, readiness for change, learning style, ability, resources and motivation. It incorporates the physical, psychological and social management of living with a chronic illness. It uses didactic and non-didactic education sessions along with social, behavioral and psychological interventions.

1.2.1 Meta-analyses of diabetes education

Several trials have been published examining the effect of DSME on clinical outcomes including glycemic control, body weight, blood pressure, lipids, and

requirement for blood glucose lowering medications (Norris et al., 2001). As most trials use glycemic control as a primary outcome, this outcome of diabetes education has been examined in meta-analyses.

A meta-analysis of 21 randomized control trials (RCTs) published between 1990 and 2000 examined 28 diabetes educational interventions on glycemic control (Ellis et al., 2004). The trials included a total of 2439 participants with the trial size ranging from 23 to 320. They included a heterogeneous group of interventions including didactic teaching, dictated goal setting, goal setting negotiated teaching method, situation problem solving, cognitive reframing interventions and other techniques with some studies incorporating more than one teaching method. The content included various combinations of

information on diet, exercise, SMBG, basic diabetes knowledge, medication adherence, and psychosocial topics. The duration and number of interventions ranged from 1 month

(12)

4 to 1 year and 1 to 36 visits. The time period to the first post intervention glycated

hemoglobin (A1C) ranged from 3 to 15 months with a net decrease in A1C of -0.320%

(95% confidence interval (CI) -0.571%, -0.069%) using the fixed effects meta-analysis (test for heterogeneity Q=14, degrees of freedom (d.f.)=19, p=0.78). The net A1C

decrease of trials with a 3 and 12 month follow up were not significant, but those with a 6 month follow up had a significant net A1C decrease of -0.486% (95% CI -0.923%, - 0.049%). There was also a significant improvement in A1C in the control group from baseline at -0.66% (95% CI -1.054, -0.265) suggesting beyond standard of care some

“control groups” also received education and increased visits.

The random effects meta-analysis of glycemic change from baseline showed a drop in A1C of -1.136% (95% CI -1.481% to -0.790%) at the end of time period 1 (Q=

132, d.f. 27, p<0.001). The change was also statistically significant at the end of 3, 6 and 12 months with change in baseline A1C -1.238% (95% CI -1.665% to -0.811%), -0.892%

(95% CI -1.428% to -0.356%) and -1.544% (95% CI -2.26% to -0.828%) respectively.

Face to face interventions, using a cognitive reframing teaching model or that included content on exercise had a larger effect on decrease in A1C.

A meta-analysis also including earlier studies published between 1980 and 1999 involving health education in diabetes mellitus with glycemic control as a primary outcome incorporated 31 studies containing a total of 4263 patients (Norris et al., 2002).

At the end of the intervention the A1C was decreased by 0.76% (95% CI 0.34 – 1.18%) more in the intervention group compared to the control group. At 1-3 months of follow up the A1C was a 0.26% (95% CI 0.21-0.73%) less in the intervention group than the control group and 0.26% (95% CI 0.05-0.48%) less in the intervention group than the control

(13)

5 group at >4 months of follow up. The meta-regression using the change in A1C as the dependent variable showed only total contact time was significant. Each additional hour of DSME reduced A1C by 0.04% (95% CI 0.01%-0.08%).

A more recent meta-analysis of DSME by Minet et al., (2010) included 47 studies published up until 2007, with its earliest included study published in 1988, involving a total of 7677 participants. Eighteen studies used behavioral psychosocial techniques which included cognitive, behavioral and motivational approaches or psychology

centered counseling in the intervention and 29 studies used educational techniques which used a didactic-oriented intervention focused on knowledge acquisition. Studies included individual and group sessions. The pooled mean difference in A1C between patients assigned to self-care management intervention was 0.36% (95% CI 0.21-0.51) compared to the control group by a random-effects model. The chi-squared for heterogeneity was significant (p<0.001). The pooled estimate with a fixed-effects model was similar at 0.30% (95% CI 0.237-0.367). The factors which may influence the effect size of A1C change were studied by meta-regressions. The univariate meta-regression found a greater reduction in A1C in studies with a follow up period ≤ 12 months (effect size 0.49%, p=0.017). Those studies with a sample size ≤ 99 had a greater reduction in A1C (effect size 0.42%, p=0.007) compared to studies with a sample size >99. The difference in A1C reduction in studies using education techniques compared to behavioural psychosocial techniques and length of intervention were not significant.

These meta-analyses showed a decrease in A1C with diabetes education

intervention. However, the degree of decrease varied between the meta-analyses and at different time points within them. Part of this difference may be due to the different time

(14)

6 periods over which the results were included or a difference in the regression models used. Ellis et al., (2004) found the effect on A1C to be related to face to face

interventions, using a cognitive reframing teaching model or that included content on exercise while Norris et al., (2002) found the effect on A1C to be related to only the total contact time and Minet et al., (2010) found the effect on A1C to be related to shorter follow up period after the intervention and small sample size.

1.2.2 Individual diabetes education

Individual diabetes education was examined in a meta-analysis by Duke et al., (2009) which included RCTs and controlled clinical trials that had at least a 6 month follow up period published until 2007. Individual education was compared to usual care in 7 studies. The 3 studies that assessed A1C at 6-9 months, included 295 participants, showed a trend toward decrease in A1C but this did not reach significance. The 4 studies involving 632 patients that examined A1C at 12 to 18 months found no significant change in glycemic control. There was a significant benefit to individual education on glycemic control in a sub analysis of 3 studies involving participants with a mean baseline A1C of

>8% with a decrease of -0.3% (95% CI -0.5 to -0.1%, p=0.007). The 2 studies that

compared individual education to group education found no significant difference in A1C at 12 to 18 months. There was no significant difference in body mass index (BMI), and blood pressure between the care types. In this meta-analysis individual care is as effective as group care and usual care for effect on A1C.

(15)

7 1.2.3 Group based diabetes education

The effect of group-based diabetes education on clinical and lifestyle outcomes has been assessed in a meta-analysis by Deakin et al. (2005). It included 11 RCTs and control clinical trials of 1532 participants of group education with groups of at least 6 participants compared to routine care, waiting list control or no intervention with follow up periods of at least 6 months. The meta-analysis of 3 studies with low heterogeneity (I²= 36.7%) shows group based diabetes education reduced A1C at 4 to 6 months by 1.4%

(95% CI 0.8-1.9%, p<0.00001) compared to the control groups. This decrease was also seen at 12 to 14 months with a reduction of 0.8% (95% CI 0.7-1.0%; p<0.00001) in the 7 studies with low heterogeneity (I²=18%) and at 2 years with a reduction of 1.0% (95% CI 0.5 to 1.4%, p<0.0001) in the 2 studies that examined it (I²=0%). Due to heterogeneity between studies, fasting blood glucose results could only be combined between 4 studies at 1 year with a reduction of 1.2 mmol/L (95% CI 0.7-1.6, p<0.0001) in favour of group education. Group education had a positive effect on body weight but no effect BMI at 12- 14 months in 4 studies with a 1.6 kg (95% CI 0.3-3.0, p=0.02) weight loss in those who received group education. There was a reduction in systolic blood pressure at 4-6 months (5 mm Hg, 95% CI 1-10, p=0.01) but the difference was not significant at 12 months. No difference in lipid profiles was found between groups at any time point. There was a reduction in the need for diabetes medications in those receiving group education with an odds ratio (OR) of 11.8 (95% CI 5.5-26.9, p<0.00001). This meta-analysis shows positive effects of group education compared to usual care on glycemic control, body weight, blood pressure and use of diabetes medications.

(16)

8 A recent meta-analysis by Steinsbekk et al. (2012) examined group based DSME compared to routine treatment. It included 21 RCTs from 1988-2007 in 26 publications with a total of 2833 participants. A1C was reduced by – 0.44% (95% CI -0.69 to -0.19%, p= 0.0006, I² =56%, 13 studies) at 6 months with group based DSME. At 12 months A1C was reduced by -0.46% (95% CI -0.74 to -0.18%, p=0.001, I²= 65%, 11 studies).

However, when 2 studies were removed due to the outlying results contributing to the high heterogeneity the 9 remaining studies showed a reduction in A1C of -0.50% (95%

CI -0.73 to -0. 27%, p<0.0001, I²=33%). In the 3 studies that followed patients to 2 years there was a -0.87% (95% CI -1.25 to -0.49%, p<0.0001, I²=0) reduction in A1C. Sub group analysis suggested that group DSME delivered by a single educator, over more than 12 hours in less than 10 months in 6 to 10 sessions gave the best improvement in glycemic control.

Group based diabetes education was found to reduce A1C in these meta-analyses with positive effects also seen on body weight, blood pressure and use of diabetes medications. Steinsbekk et al. (2012) found that group DSME delivered by a single educator, over more than 12 hours in less than 10 months in 6 to 10 sessions, gave the best improvement in glycemic control.

1.2.4 Effect of diabetes education on cardiovascular risk factors

The Steinbekk et al. (2012) meta-analysis also examined various lifestyle outcomes and cardiac risk factors. There was a significant improvement in diabetes knowledge and self-management skills with group DSME but the heterogeneity was high.

Self-efficacy and empowerment was significantly increased with a standard mean

(17)

9 difference of 0.28 (95% CI 0.06 to 0.50, p=0.012, I²=0). Treatment satisfaction was increased at 6 and 12 months with a standard difference in the mean of 0.65 (95% CI 0.44 to 0.85, p<0.0001, I²=0, 2 studies) and 0.39 (95% CI 0.21 to 0.57, p<0.0001, I²=0.0, 3 studies). However, there was no difference in quality of life. Weight was significantly improved at 12 months but not at 6 months. There was no significant difference in mortality, BMI, blood pressure and lipid profile.

A systematic review of diabetes education trials published between 1980 and 1999 found several studies that examined cardiovascular risk factors (Norris et al., 2001).

Thirteen studies had a positive effect on weight loss, while many did not. The studies with a positive effect generally involved regular contacts or reinforcement sessions or a short follow up period with those studies of ≥6 months follow up having no significant difference between groups. Some studies found an improvement in total cholesterol, low density lipoprotein (LDL) and high density lipoprotein (HDL) cholesterol with self- management training where others found an initial positive result but no significant difference at the final follow up. The individualized and repetitive interventions were more likely to improve lipid levels where as didactic interventions did not improve lipid profiles.

1.2.5 Recent trials of diabetes education

Some more recent studies on DSME have been published that were not included in the earlier meta-analysis and systematic reviews. They include DESMOND (Diabetes Education and Self-management for Ongoing and Newly Diagnosed type 2) from the United Kingdom (Davies et al., 2008) and Rethink Organization to iMprove Education

(18)

10 and Outcomes (ROMEO) from Italy (Trento et al., 2010) as well as studies from

Germany, the Netherlands and Malaysia. These studies instituted a variety of DSME interventions as outlined and examined the effect on glycemic control as well as cardiovascular risk factors.

DESMOND, a multicentre cluster RCT of 824 patients at 207 general practices in 13 primary care sites in United Kingdom compared a structured education program delivered by 2 healthcare professional educators to usual care (Davies et al., 2008). The program consisted of 6 hours delivered in 1 day or 2 half days using a non-didactic approach to present the curriculum focused on lifestyle factors as part of self-

management. DESMOND showed no difference in the primary outcome of A1C at 12 months with a non-significant decrease in both group education and usual care. Weight loss was greater in the group education setting at −2.98 kg (95% CI −3.54 to −2.41 kg) compared with −1.86 kg (95% CI −2.44 to −1.28 kg), (p=0.027) at 12 months. There was no difference in cholesterol profile, blood pressure or waist circumference between groups. The OR for using an oral hypoglycaemic agent at month 12 of the trial was 0.79 (Gillett et al., 2010). The OR of using a statin at month 12 of the trial was 0.99. The OR of antihypertensive use was 1.18 (95% CI 0.71 to 1.98). Smoking was higher in the control group compared to the intervention group at 12 months with an OR of 3.56 (95%

CI 1.11 to 11.45, p=0.033) (Davies et al., 2008). The intervention group had a significantly improved understanding of diabetes with a greater improvement in the illness belief scores and lower depression score than the control group. The estimated mean incremental lifetime cost per person receiving the DESMOND intervention was

£209 (95% CI −£704 to £1137) (Gillett et al., 2010). The incremental gain in quality

(19)

11 adjusted life years (QALY) per person is 0.0392 (95% CI −0.0813-0.1786) with a mean incremental cost per QALY of £5387. Using simulated long term effect there is 66%

likelihood that the DESMOND intervention is cost effective.

A RCT of patients with non-insulin dependent type 2 diabetes managed by systemic group education or usual care of individual consultations and education was performed in 112 patients in Italy (Trento et al., 2001) with follow up at two, four (Trento et al., 2002) and five years (Trento et al., 2004). At 2 years the A1C had increased in the control group from 7.4% ± 1.4% to 8.3% ± 1.8% but remained unchanged in the

intervention group at 7.4% ± 1.4% to 7.5% ± 1.4% (p<0.002). HDL cholesterol was increased in group patients but not control patients. There was no significant difference in fasting blood glucose, total cholesterol, triglycerides, creatinine, albuminuria, body weight, BMI, foot ulcers or diabetic retinopathy between groups. Knowledge of diabetes, health behaviors and quality of life were also significantly improved with group care. At four years, A1C remained stable in the group patients from 7.4% ± 1.4% at baseline to 7.0

± 1.1% and continued to rise in the individual care patients from 7.4% ± 1.4% at baseline to 8.6% ± 2.1% (p<0.001) (Trento et al., 2002). Body weight and BMI decreased and HDL increased in the group patients over 4 years but there was no significant difference in the individual care patients. The knowledge of diabetes, health behaviours and quality of life improved in the group care patients and declined in the individual care group.

Diabetes medication use was decreased in the group care patients compared to the individual care patients suggesting improved compliance to lifestyle modification with diabetes education leading to better control of diabetes (Trento et al., 2002). At five years follow up A1C in the group patients continued to remain stable at 7.3% ± 1.0% compared

(20)

12 to an increase in control patients to 9.0% ± 1.6% (p<0.001) (Trento et al., 2004). At five years knowledge of diabetes, problem solving ability, and quality of life continued to improve in the group patients and decrease in the control patients. Body weight but not BMI was decreased from baseline in the group patients at five years. The group patients maintained better long-term glycemic control than the individual care patients suggesting group care may be more effective.

The ROMEO trial is a multicenter trial performed to examine if the results of group care model of diabetes lifestyle intervention from a single center described by Trento et al. (2004) was transferable to other clinics (Trento et al., 2010). It was a 4 year RCT of group care vs. routine individual care of 815 patients with non-insulin dependent type 2 diabetes performed in 13 diabetes clinics in Italy. At 4 years the group patients had a lower A1C at 7.3% ± 0.9% compared to 8.8% ± 1.2% in the routine care patients

(p<0.001) and the OR of having an A1C ≤ 7.0% was 29.4 (95% CI 14.2-60.8, p<0.001) in favour of group care. Group care subjects also had significantly higher HDL cholesterol, lower fasting glucose, LDL cholesterol, triglycerides, blood pressure, body weight, BMI, and creatinine compared to control subjects. At the end of the study, group care patients were more likely to have reached all treatment targets whereas fewer control patients had an A1C ≤ 7.0% at study end compared to at the initiation and there was no change in the proportion reaching the other treatment targets. Health behaviours, quality of life and diabetes knowledge were significantly improved for group care patients compared to no change in health behaviours and a worsening of quality of life and knowledge seen in control patients. Prescriptions for hypoglycemic, antihypertensive and lipid lowering medications were similar between groups suggesting healthier behaviors with group care.

(21)

13 ROMEO demonstrates that continuing interactive patient-centered education by group care significantly improves outcomes.

A prospective RCT of three education programs for 181 type 2 diabetes patients in Germany compared a didactic-oriented training program, a self-management-oriented program delivered in a group setting and a more individualized self-management program where half the sessions were conducted in an individual setting (Kulzer et al., 2007). The didactic program focused on the knowledge, skills and information about the treatment of diabetes in four 90 minute lessons in a group setting in a program which has been

previously studied and used since the late 1980’s. The self-management program focused on the emotional, cognitive and motivational process of behavior change and was

delivered twelve 90 minute lesions in a group setting. The final program had the same content of the second program but was delivered in 6 group sessions and 6 individual sessions. The group based self-management program had a 0.7% fall in A1C that was sustained at 12 months after completion of the intervention (p=0.013). There was no change in A1C in the didactic program. The individualized program had a drop in A1C at 3 months but this was not maintained at 12 months. There were also benefits seen for BMI, fasting blood glucose, psychological variables and exercise in the group self- management program.

A RCT of 54 patients in the Netherlands with type 2 diabetes treated with maximum oral agents with an A1C≥7.0% was performed to examine the long term outcome of DSME (Goudswaard et al., 2004). Patients were randomly assigned to 6 month education program by diabetes nurse or usual care by their general practitioner with a primary outcome of the proportion of patients with an A1C<7% 1 year after the

(22)

14 end of the intervention. An A1C < 7% was achieved by 60% of the patients in the

diabetes education group and by 17% of the patients in the usual care group (OR 6.6;

95% CI 1.8-24.5, p=0.004). The mean A1C in the diabetic education group fell from 8.2% ± 1.1% to 7.2% ± 1.3% and in the usual care group from 8.8% ± 1.5% to 8.4% ± 1.7%. The difference in the mean change in A1C was not statistically significant at 0.2%

(95% CI -0.7% to +0.4%) in favor of the diabetes education group when adjusted for baseline values.

A RCT of a 12 week DSME program in Malaysia consisting of two in-person individual education sessions and one telephone follow up was compared to usual care in 164 patients (Tan et al., 2011). At the end of the 12 week intervention there was a

significant reduction in total daily calorie intake and increase in activity in the

intervention group based on self-reported food diaries and questionnaires. Based on self- reported questionnaires, 91% (95% CI 89-94%) of the intervention group were adherent to prescribed medications (defined as consuming ≥ 90% of prescribed medications in the previous week) compared to 84% (95% CI 82-87%) (p=0.008) of the usual care group at week 12. There was also more SMBG in the intervention group at the end of 12 weeks compared to no change in the control group based on the count of returned glucose test strips and self-monitoring diaries. At 12 weeks the A1C was lower in the intervention group at 8.75% ± 1.75% compared to 9.67% ± 2.01% in the control group (p<0.001). The A1C difference at 12 weeks persisted after adjusting for medication adherence, SMBG frequency and body weight.

These more recent studies of DSME had variable effects on glycemic control and other outcomes. Trento et al. (2001, 2002, 2004) had shown improvements in glycemic

(23)

15 control and other outcomes with patient centered group diabetes education that were confirmed in the ROMEO trial (Trento et al., 2010). DESMOND, did not find a

significant difference in the decrease in A1C between groups (Davies et al., 2008). It did find a significant benefit with the structured diabetes education program with weight loss and smoking cessation. The smaller studies did show a significant difference in A1C however one study had only a 12 week follow up. These studies also showed benefit in other outcomes such as BMI and other self-management behaviors. These recent studies together with the meta-analysis suggest a positive effect of DSME on various outcomes.

1.2.6 Methodological issues with diabetes education trials

There are many methodological issues with these trials examining diabetes education (Norris et al., 2001). Descriptive information is frequently lacking in many trials, including details of the study population such as the type of diabetes, as well as details of interventions. The usual care of the control groups in each study varies and is not always defined. The study populations may not be representative of the target population due to selection, performance, and attrition bias (Juni et al., 2001). Selection bias may exist when there are systemic differences in the control and intervention groups at baseline. This possibility may only be excluded with randomization. The

generalizability of the results may be limited by provider and patient selection simply due to their willingness to participate in a trial (participation bias). Differences in provider and patient behavior can also result from the Hawthorne effect, i.e. subjects that know they are part of a study behave differently due to this fact alone. These studies were often performed at tertiary care centres or university hospitals, which may differ from

(24)

16 community settings in the types of patients and the care they receive. Some studies

limited enrollment to newly diagnosed patients with type 2 diabetes which limits the generalizability to the larger population with diabetes. Performance bias may result from differences in care provided to the control and intervention groups other than the

intervention being evaluated. To prevent this there should be no evidence of

contamination or co-intervention, including no additional contacts with researcher or providers for the intervention group compared with the control group or compared to routine care. Additional clinical resources, intensity of follow-up, and other factors related to a study can make all trial participants (regardless of randomization arm) different from the real world patients in clinical care. Attrition bias results from different rates of withdrawal from the study between groups. To avoid this, the attrition rates should generally be <20% of the total number and dropouts must resemble completers in baseline characteristics. These numerous factors that make the study population different from the general population of real-world clinical care can threaten the external validity of the studies of DSME.

In studies of diabetes education, the internal validity was threatened by a variety of factors. The assessors were often not blinded and it is impossible to blind the study subjects. There was a lack of information on the process of randomization and allocation concealment. There were high attrition rates in some studies and evidence of co-

interventions with some control groups being more frequently than standard of care and receiving some form of education. There was a potential for response-set bias where the intervention group self-reported dietary, exercise and glucose self-monitoring habits that

(25)

17 match the goals of the intervention rather than actual behavior. The instruments used to measure diabetes knowledge, self-care and dietary habits, have not been validated.

The actual DSME intervention also takes many forms, including individual

sessions, group sessions, didactic teaching and cognitive behaviour therapy. Many studies included fewer than 100 participants and few studies included a statistical power

calculation. While these interventions may be efficacious in selected study populations under the supervision of study investigators, they may not be as effective in the general population where there is a wider variety of clinicians and patients. There is little real world evidence of the effectiveness of diabetes education.

1.3 Ontario Diabetes Education Centres

A survey with linked health care administrative data of 781 patients with diabetes greater than 2 years duration in Ontario in 2003-2004 examined predictors of Diabetes Education Centre (DEC) attendance and quality of care indicators for effectiveness of DSME (Shah et al., 2009). Of the respondents, 30% had attended a DEC in 2002.

Predictors of DEC attendance included recently diagnosed diabetes, receiving regular specialist care, receiving regular primary care visits and marital status. A propensity score model derived from demographics, health service utilization, diabetes clinical features, and other medical conditions was used to examine quality of care indicators such as capillary glucose testing, retinal screening examination, acute diabetes complications and continuity of primary care between the attendees and non-attendees. DEC attendees were more likely to receive retinal screening examination in the 2 years following than those who did not attend a DEC. There was no difference in the other quality of care indicators.

(26)

18 Attendance at DECs in Ontario has previously been shown to be associated with more glucose self-monitoring in attendees vs. non-attendees (86.2% vs. 53.5%, adjusted OR 6.45, 95% CI 5.61-7.42, p < 0.0001) in a cross-sectional study of a subset of the over 65 population in Ontario (Millar et al., 2010). Anti-hypertensive medications (88.3% vs.

83.0%, adjusted OR 1.30, 95% CI 1.12–1.51, p=0.0006) and lipid-lowering drugs (77.2%

vs. 65.6%, adjusted OR 1.68, 95% CI 1.50–1.88, p <0.0001) were also used more by the DEC attendees. DEC attendees were also more likely to have an eye examination (adjusted OR 1.13, 95%CI 1.02–1.26, p=0.0229). This study showed an association between DEC attendance and objective evidence of better management of diabetes.

A population based cohort study of all adults with diabetes who attended DECs in Ontario in 2006 examined DEC attendance by those with newly diagnosed diabetes (Cauch-Dudek et al., 2013). Only 20.6% of those with newly diagnosed diabetes attended a DEC within 6 months of diagnosis. Patients of older age, lower socioeconomic status and recent immigrants were less likely to attend. Mental health conditions and other medical comorbidities were also associated with not attending DECs suggesting those most in need of DSME are not receiving this resource.

1.4 Diabetes guidelines

Many studies have shown the benefits of glycemic control, antihypertensive medications and cholesterol lowering medications in lowering morbidity and mortality in diabetes. The targets of the Canadian Diabetes Association Guidelines are based on this evidence (Cheng et al., 2013). The recommended glycemic target is an A1C of ≤7.0% to reduce microvascular complications and, in type 1 diabetes, macrovascular complications

(27)

19 (Imran et al., 2013). In Canada, males with diabetes greater than 45 years of age and females with diabetes greater than 50 years of age are considered to be at high risk for cardiovascular disease and should be considered for vascular protection (Stone et al., 2013). This includes optimization of blood pressure to the target of <130/80 mmHg which often requires multiple antihypertensive agents, starting with angiotensin

converting enzyme (ACE) inhibitor or angiotensin receptor blocker (ARB) therapy, and lipid-lowering medication, primarily statins, to target an LDL cholesterol of ≤ 2.0 mmol/L. All patients with type 2 diabetes and patients with type 1 diabetes greater than 15 years of age should undergo screening for retinopathy every 1 to 2 years (Boyd et al., 2013). The information and recommendations in the guidelines, based on the evidence for glycemic control, lipid and blood pressure management, complication screening and management as outlined below provide a basis for the goals of DSME.

1.5 Benefits of glycemic control 1.5.1 Type 1 diabetes

The Diabetes Control and Complications Trial (DCCT) randomized 1441 patients with type 1 diabetes, to intensive (goal A1C<6.0%) or conventional therapy for a mean of 6.5 years between 1983 and 1993 (DCCT group, 1993). At the end of the study the difference in A1C was 1.7% (7.4% in the intensive-treatment group vs. 9.1% in the conventional-treatment group, p<0.01). The intensive-treatment group had a 76% risk reduction (95% CI 62 to 85%, p<0.001) for the development of retinopathycompared to conventional therapy. With intensive therapy there was also slowed progression of pre- existing retinopathy with a 54% risk reduction (95% CI 39 to 66%, p<0.001). The risk reduction in the development of proliferative or severe non-proliferativeretinopathy was

(28)

20 47% (95% CI 14to 67%, p=0.011). The absolute risk reduction in the development of severe nephropathy was 4.04% in the intensive group. The risk of microalbuminuria was reduced by 39% (95% CI 21 to 52%, p≤0.002) and albuminuriaby 54%(95% CI 19 to 74%, p<0.04) in the intensive group. The risk of clinical neuropathy at 5 years was also reduced by 60% (95% CI 38 to 74%, p≤0.002). The absolute risk reduction in the development of clinical neuropathy in the primary prevention group was 7% (p=0.006) with intensive treatment and 9% in the secondary prevention cohort (p<0.001). During DCCT, there were fewer cardiovascular events in the intensive group but the young age of the cohort and small number events did not lead to a statistically significant difference.

The DCCT showed that tight glycemic control with an A1C of 7.4% in the intensive treatment group is associated with a reduced risk of neuropathy, nephropathy and retinopathy in type 1 diabetes.

Ninety-three percent (n=1397) of participants were followed until February 1, 2005, during the observational Epidemiology of Diabetes Interventions and

Complications (EDIC) study (Nathan et al., 2005). At the end of DCCT, the conventional group was advised to follow intensive control and the intensive group went back to regular clinical care. At the end of the EDIC study period there was no difference between groups in A1C, 7.9% in the intensive group and 7.8% in the conventional group.

The time to first to cardiovascular event became statistically significant during the follow up observational period, suggesting a lasting benefit of tight control as the 2 groups had no difference in A1C during this period (Nathan et al., 2005). The intensive arm had a 42% risk reduction (95% CI 9-63%; p=0.02) compared to the control group at a mean 17 years of follow up, 7 years after the study intervention was completed. The event rates

(29)

21 for the first cardiovascular event were 0.38 and 0.80 per 100 patient-years in the intensive and control group respectively (p=0.007). The risk reduction for first occurrence of nonfatal myocardial infarction (MI), stroke, or death from cardiovascular disease was even greater at 57% (95% CI 12-79%; p=0.02). The effects on microvascular disease were also preserved in follow up (Writing team for DCCT/EDIC Research Group, 2002).

EDIC demonstrated that early intensive glycemic control is associated with a reduction in both microvascular and macrovascular disease in patients with type 1 diabetes.

1.5.2 Type 2 Diabetes

The United Kingdom Prospective Diabetes Study (UKPDS) examined intensive glucose control to conventional therapy in newly diagnosed type 2 diabetes (UKPDS 33 and 34, 1998). The participants were followed for an average of 10 years from 1977- 1997. The average A1C was 7.0% in the intensive group vs. 7.9% in the control group in the sulfonylurea/insulin study (UKPDS 33, 1998) and A1C 7.4% vs. 8.0% in the

metformin study (UKPDS 34, 1998). The relative risk of any diabetes-related end point was 0.88 (95% CI 0.79-0.99, p=0.029) in the intensive treatment group with sulfonylurea and insulin (UKPDS 33, 1998). The relative risk of microvascular disease was 0.75 in the sulfonylurea/insulin arm (95% CI 0.60-0.93, p=0.0099). This was mainly driven by the reduction in the need for retinal photocoagulation and cataract extraction. There was no difference in diabetes related death mortality, all-cause mortality or macrovascular disease. UKPDS showed that tight glycemic control with an A1C of 7% early in diagnosis of type 2 diabetes reduces the risk of microvascular disease.

(30)

22 78% of the participants in UKPDS agreed to enroll in post-trial monitoring and were followed for 10 more years, the predicted 50% mortality rate point (Holman et al., 2008). The difference in A1C between groups was lost after the first year following the completion of the study intervention. At 10 years post study, there was a significant decrease in any diabetes related end point in the patients originally in the

insulin/sulfonylurea group with a 9% relative risk reduction (p=0.04). Diabetes related death was reduced by 17% (p=0.01), death from any cause by 13% (p=0.007), MI by 15%

(p=0.01), and microvascular disease by 24% (p=0.001). In the patients originally in the metformin treated group, the reduction in any diabetes related end point persisted at 10 years post study with a relative risk reduction of 21% (p=0.01). The difference in diabetes related death, MI and all-cause mortality with metformin persisted at 10 years with relative risk reductions of 30% (p=0.01), 33% (p=0.005) and 27% (p=0.002) respectively.

The intensive glycemic control early in type 2 diabetes continued to provide benefit in macrovascular and microvascular disease 10 years after cessation of the trial despite a loss of difference in A1C between groups suggesting a legacy effect of tight glucose control (Chalmers et al., 2008).

ACCORD (Action to Control Cardiovascular Risk in Diabetes Study) randomized 10,251 patients with a median A1C 8.1% and 10 year duration of type 2 diabetes to intensive glucose control with a target A1C of <6.0% or standard therapy (Gerstein et al., 2008). An A1C of 6.4% was achieved in the intensive group compared to 7.5% in the control group. The primary outcome, a composite end point of nonfatal MI, stroke, or death from cardiovascular disease, had a hazard ratio (HR) of 0.90 (95% CI 0.78-1.04;

p=0.16) when the trial was stopped at 3.5 years due to an increase in death in the

(31)

23 intervention group. There was a 22% increase in mortality with a 35% increase in

cardiovascular mortality with a HR of 1.22 for all-cause mortality (95% CI 1.01-1.46;

p=0.04) and a HR of 1.35 for cardiovascular mortality (95% CI 1.04–1.76; p=0.02) with intensive treatment. Hypoglycemia was significantly higher in the intensive group. Post hoc analysis of episodes of severe hypoglycaemia and differences in the use of drugs including rosiglitazone, weight change, and other factors could not explain the increased mortality with intensive therapy. ACCORD provides evidence that older patients with long standing type 2 diabetes may not benefit from tight glycemic control with an A1C

<6%.

ADVANCE (Action in Diabetes and Vascular Disease: Preterax and Diamicron Modified Release Controlled Evaluation) randomized 11,140 patients with an 8 year mean duration of diabetes, a mean A1C 7.5%, and 32% had a history of macrovascular disease, to standard glucose control or intensive glucose control using gliclazide MR plus other drugs as required to achieve A1C ≤ 6.5% (Patel et al., 2008). During the 5 year study the groups obtained an average A1C 6.5% and 7.3% respectively. The occurrence of combined major macrovascular and microvascular events was reduced by intensive control with a HR of 0.90 (95% CI 0.82 to 0.98; p=0.01). Major microvascular events were also reduced by intensive control with a HR of 0.86 (95% CI 0.77 to 0.97; p=0.01).

This was driven by a reduction in the incidence of nephropathy (HR 0.79; 95% CI 0.66 to 0.93; p=0.006). ADVANCE demonstrated that an A1C of 6.5% in patients with type 2 diabetes of 8 years duration is associated with a reduction in nephropathy but did not affect macrovascular disease and other microvascular disease.

(32)

24 Veteran’s Affairs Diabetes Trial (VADT) randomized 1791 military veterans with a 11.5 year duration of type 2 diabetes, 40% incidence of a prior cardiovascular event, and a baseline A1C 9.4% to intensive control with an A1C of <6% vs. standard control (Duckworth et al., 2009). They achieved an A1C of 6.9% in the intensive group vs. 8.4%

in the standard group with a mean follow up of 5.6 years. There was no difference in the primary outcome of major cardiovascular events, in any component of the primary outcome, death from any cause, or microvascular complications.

An A1C of <7% has been shown to reduce microvascular complications of type 1 and type 2 diabetes in multiple trials (DCCT group, 1993; UKPDS 33, 1998; UKPDS 34, 1998; Patel et al., 2008). More intensive glucose control has not shown any significant reduction of cardiovascular disease compared to standard glycemic control during the randomized portion of the trials (DCCT group 1993; UKPDS 33, 1998; Gerstein et al., 2008; Patel et al., 2008; Duckworth et al., 2009) despite the association of high A1C with cardiovascular disease. The post-trial follow-up periods of DCCT/EDIC and UKPDS did show a reduction in cardiovascular disease suggesting a legacy effect of early glycemic control (Nathan et al., 2005; Holman et al., 2008).

1.6 Multifactorial intervention

STENO2 randomized 160 patients with type 2 diabetes and microalbuminuria to intensified, multifactorial intervention or conventional treatment in accordance with national guidelines on risk factors for cardiovascular disease in an open label parallel trial conducted in Denmark from 1993-2001 (Gaede et al., 2003). The intensive treatment consisted of stepwise implementation of behaviour modification and pharmacologic

(33)

25 therapy to target hyperglycemia to A1C <6.5%, hypertension, dyslipidemia,

microalbuminuria, smoking cessation and secondary prevention of cardiovascular disease with aspirin. The A1C at the end of the study period was 9.0% ± 1.8% in the conventional group and 7.9% ± 1.2% (p <0.001) in the intensive group with about 15% obtaining the goal A1C <6.5%. The risk of cardiovascular disease was significantly reduced with intensive therapy with a HR of 0.47 (95% CI 0.24 to 0.73, p=0.008). Nephropathy (HR 0.39; 95% CI 0.17 to 0.87), retinopathy (HR 0.42; 95% CI 0.21 to 0.86), and autonomic neuropathy (HR 0.37; 95% CI 0.18 to 0.79) were also reduced in the intensive

intervention group.

STENO2 patients were then followed for 5.5 years after the end of the study to a total of 13.3 years follow up (Gaede et al., 2008). By the end of the follow up period there was no difference in A1C between the groups. The risk of death from any cause was decreased with intensive therapy (HR 0.54; 95% CI 0.32 to 0.89; p=0.02). Intensive therapy also lowered the risk of death from cardiovascular causes, cardiovascular events, and end-stage renal disease, and the need for retinal photocoagulation.

STENO2, which achieved an A1C 7.9% as well as targeting multiple risk factors, significantly lowered the risk of cardiovascular disease (Gaede et al., 2003) with the effect preserved in post study follow up (Gaede et al., 2008). Nephropathy, retinopathy and autonomic neuropathy were also reduced in the intensive intervention group (Gaede et al., 2003) with a reduction in progression to end stage renal disease and need for retinal photocoagulation seen in the post study follow-up (Gaede et al., 2008). STENO2 program of behavioural modification and pharmacologic therapy to target a multiple risk factors requires a multidisciplinary approach that would include diabetes education to implement

(34)

26 the required teaching and behaviour modification to achieve outcomes that are not be achieved with glycemic control alone.

1.7 Lipid management in Diabetes

The Heart Protection Study (HPS) diabetes cohort of 5963 subjects showed treatment with 40 mg simvastatin daily resulted in a 27% reduction in cardiovascular events and a 25% reduction in stroke relative to placebo, which was similar to the cohort without diabetes (Collins et al., 2003). The Collaborative Atorvastatin Diabetes Study (CARDS) was conducted in people with type 2 diabetes with a mean baseline LDL of 3.1 mmol/L without known vascular disease and at least 1 additional risk factor for

cardiovascular disease (Colhoun et al., 2004). Treatment with atorvastatin 10 mg daily to a mean LDL of 2.0 mmol/L reduced the risk of first cardiovascular disease events by 37%

and risk of stoke by 48%.

The Anglo-Scandinavian Cardiac Outcomes Trial – Lipid Lowering Arm

(ASCOT-LLA) of 10,305 hypertensive patients with no history of coronary heart disease but at least three cardiovascular risk factors were randomly assigned to receive 10 mg atorvastatin or placebo for a mean follow up of 3.3 years (Server et al., 2005). The subgroup analysis of 2,532 patients with type 2 diabetes at the time of randomization showed similar benefit of atorvastatin as seen in the entire cohort. The atorvastatin group had 116 (9.2%) major cardiovascular events or procedures compared to 151 (11.9%) events in the placebo group (HR 0.77, 95% CI 0.61– 0.98; p=0.036). The number of events occurring in the diabetes subgroup was small and although there were less coronary events (HR 0.84, 95% CI 0.55–1.29; p=0.14) and strokes (HR0.67, 95% CI

(35)

27 0.41–1.09; p=0.66) in the atorvastatin group, the reductions were not statistically

significant.

The Treating to New Targets (TNT) trial included a diabetic subgroup of 1051 subjects with stable coronary artery disease treated with atorvastatin 80 mg or 10 mg daily (Shepherd et al., 2006). Subjects treated with atorvastatin 80 mg daily that achieved a group mean LDL of 2.0 mmol/l had 25% fewer major cardiovascular events than those treated with atorvastatin 10 mg daily who achieved a mean LDL of 2.5 mmol/L

(p=0.026). Atorvastatin 80 mg also reduced the rates of all cardiovascular and

cerebrovascular events compared to atorvastatin 10 mg daily. The diabetes subgroup had an increased event rate for all primary and secondary efficacy outcomes compared to the overall study population. This reinforces the evidence that people with diabetes and coronary artery disease are at high risk of subsequent cardiovascular events.

The evidence from HPS, CARDS, ASCOTT-LLA and TNT show the benefit of statin treatment for LDL cholesterol in patients with diabetes at risk for cardiovascular and cerebrovascular disease. This important risk factor for macrovascular disease is one of the topics addressed in diabetes education.

1.8 Blood Pressure management

UKPDS showed that tight blood pressure control with a mean blood pressure of 144/82 mm Hg over 8.4 years of follow up compared to 154/87 mm Hg in the

conventional arm reduced the risk of microvascular disease stroke and deaths related to diabetes (Adler et al., 2000). For every 10 mm Hg decrease in mean systolic blood pressure there was a 12% risk reduction in any complication related to diabetes (95% CI

(36)

28 10-14%, p<0.0001), a 15% risk reduction in deaths related to diabetes (95% CI 12-18%, p<0.0001), a 11% risk reduction in MI (95% CI 7-14%, p<0.0001, and a 13% risk reduction in microvascular complications (95% CI 10-16%, p<0.0001).

The Hypertension Optimal Treatment (HOT) Study of 18 790 patients, aged 50-80 years with hypertension and diastolic blood pressure between 100 mm Hg and 115 mm Hg, assigned patients to 3 groups based on target diastolic blood pressure of ≤90 mm Hg,

≤85 mm Hg and ≤80 mm Hg (Hansson et al., 1998). In the diabetes subgroup there was a 51% reduction in major cardiovascular events in the ≤80 mm Hg group compared with the ≤90 mm Hg group (p=0.005).

The Appropriate Blood Pressure Control in Diabetes (ABCD) trial of

normotensive patients with type 2 diabetes to intensive control of 10 mm Hg below the baseline diastolic blood pressure or moderate control with a diastolic blood pressure of 80-89 mm Hg (Schrier et al., 2002). Over a period of 5.3 years the average blood pressure in the intensive group was 128±0.8/75±0.3 mm Hg compared to 137±0.7/81±0.3 mm Hg in the moderate group (p<0.0001). There was no difference in the primary end point of change in creatinine clearance (p=0.43). Fewer patients in the intensive group reached the secondary end points of progression from normoalbuminuria to microalbuminuria

(p=0.012) and microalbuminuria to overt albuminuria (p=0.028). The intensive control group also has less progression of diabetic retinopathy (p=0.019) and a lower rate of strokes (p=0.03).

The ADVANCE trial examined the effects of fixed combination of perindopril and indapamide compared to placebo on macrovascular and microvascular outcomes in 11 140 patients with type 2 diabetes mellitus and hypertension (Patel et al., 2007). The

(37)

29 primary end point was a composite of major macrovascular and microvascular events.

After 4.3 years of follow up the active therapy group had a mean reduction in systolic blood pressure of 5.6 mm Hg and diastolic blood pressure of 2.2 mm Hg compared to the placebo group. The primary endpoint of major macrovascular or microvascular event had a relative risk reduction of 9% (HR 0.91, 95% CI 0.83–1.00, p=0.04). Death from

cardiovascular disease was also reduced by 18% (HR 0.82, 95% CI 0.68–0.98, p=0.03) and death from any cause was reduced by 14% (HR 0.86, 95% CI 0.75–0.98, p=0.03).

The ACCORD trial randomized 4733 people with diabetes to systolic BP targets of <120 or <140 mm Hg for a mean follow up of 4.7 years (Cushman et al., 2010). The composite primary outcome of nonfatal MI, nonfatal stroke, or death from cardiovascular causes was not significant with an annual rate of 1.87% in the intensive therapy group and 2.09% in the standard therapy group (HR 0.88; 95% CI 0.73 to 1.06; p=0.20). The annual rate of deaths from any cause was also not significant. The annual rate of stroke was decreased with intensive therapy (HR 0.59; 95% CI 0.39 to 0.89; p=0.01).

ACE inhibitors and ARBs have been shown to reduce all-cause mortality, cardiovascular events and cardiovascular mortality in patients with diabetes (HOPE investigators 2000 and Lindholm et al., 2002). ACE inhibitors and ARBs have also been shown to reduce progression of kidney disease in patients with diabetic nephropathy (Lewis et al., 1993 and Lewis et al., 2001). For these reasons they are first line therapy for hypertension in patients with diabetes (Gilbert et al., 2013).

(38)

30 1.9 Achieving metabolic targets

Despite the evidence and clinical practice guidelines, many patients do not reach metabolic targets. In the United Sates, the National Health and Nutrition Examination Survey (NHANES) 2009-2010 health survey showed only 55.5% of participants with diabetes achieved the target goal of A1C <7.0% and similar rates of compliance were seen for blood pressure and lipid recommendations (Wong et al., 2013). The blood pressure target <130/80 mm Hg was achieved by 52.8% and LDL cholesterol of <2.6 mmol/l was achieved by 54.7% of patients with diabetes. The proportion of the NHANES 2009-2010 population with diabetes who met all three targets was 24.9%. However, there is a trend over time from 1999 to 2010 for an improvement in the proportion of the NHANES population reaching these targets.

A similar situation exists in Canada with a large proportion of the population with diabetes not meeting targets for glycemic, blood pressure and lipid control. In a chart audit of family practices in Ontario the average A1C was 7.9% with 25.7% of patients having an A1C at target of <7% and 31.5% had an A1C above 8.4% (Harris et al., 2003).

In a more recent cross sectional chart audit of 243 primary care physicians in Canada the average A1C was 7.3% with 51% of patients achieving a target A1C of <7% (Harris et al., 2005). The proportion of patients with optimal control deteriorated over time from diagnosis from 69% with an A1C <7% in the first 2 years of diagnosis to 38% at target A1C at 15 or more years since diagnosis. In a cross sectional study of primary care physicians in 3 Canadian provinces (Nova Scotia, New Brunswick and Prince Edward Island) found 54% of patients with diabetes were at target for a blood pressure of <130/80 (Putnam et al., 2011). A cross sectional study of centers in Canada and Europe found that

(39)

31 41.6% of patients with diabetes already on a statin were not at target for LDL (Leiter et al., 2011).

1.10 Medication utilization

A previous study of the clinical and demographic characteristics of patients receiving oral anti hyperglycaemic medications in Ontario residents aged 66 years and older on March 31, 2008 identified 387 778 people with diabetes out of 1 752 000 Ontarians aged 66 years and older (22.1%) (Gomes et al., 2009). The mean duration of diabetes was 9.3 years with a standard deviation of 5.1 years. Most individuals (60.1%) had at least one prescription for an oral hypoglycaemic agent in the previous year.

A retrospective cohort study of newly treated elderly hypertensives in Ontario was performed using health administrative databases (Friedman et al., 2010). The diabetes cohort of 41 236 patients aged 66 years or more in Ontario between 1997 and 2005, 76.2% were prescribed an ACE inhibitor (n=31 414), 4.9% an ARB (n=2041), 4.7% a beta blocker (n=1935), 5.7% a calcium channel blocker (n=2351), and 8.5% a diuretic (n=3495).

A population based study of 105 715 people ≥65 years with newly diagnosed diabetes in Ontario between 1994 and 2001 used administrative databases to examine the receipt of antihypertensive and lipid-lowering drugs by patients cared for by different physician specialties (Shah et al., 2006). The unadjusted lipid lowering prescription rates with newly diagnosed diabetes ≥65 years was 23.9% for by those cared for family physicians, 28.6% for those cared for by internists and geriatricians and 36.1% by

endocrinologists. Of the types of lipid lowering medications prescribed, statins accounted

(40)

32 for 86.8 to 89.9%. The unadjusted antihypertensive medication prescribing rates were 66.0% for those cared for by family physicians, 75.1% of those cared for by internists and geriatricians and 69.3% of those cared for by endocrinologists. Of these antihypertensive medications, 54.2 to 63.1% were ACE inhibitors or ARBs. This study illustrates the low rates of the use of vascular protection medications and the differences in prescribing among physician specialties.

The administrative databases of Saskatchewan were used to examine the use of anti-platelet agents, statins, and ACE inhibitors by people with type 2 diabetes with and without symptomatic atherosclerosis (Brown et al., 2004). The cohort of 12 106 people with type 2 diabetes had an average age of 64 years, 55% male, with a mean duration of follow up of 5 years. Those patients with type 2 diabetes and coronary artery disease were more likely to receive antiplatelet agents at 37% compared to 15% of type 2 diabetes patients without coronary artery disease (p <0.001). They were also more likely to receive a statin at 29% vs. 15% of patients with type 2 diabetes without coronary artery disease (p<0.001) and ACE inhibitors at 60% compared to 43% (p<0.001). Patients with cerebrovascular disease and type 2 diabetes were more likely to receive an antiplatelet agent at 46% compared to 20% of patients with type 2 diabetes without cerebrovascular disease (p<0.001) and ACE inhibitors at 58% compared to 47% (p<0.001) but less likely to receive a statin at 16% compared to 20% of those with type 2 diabetes without

cerebrovascular disease (p=0.001). Patients with peripheral arterial disease and type 2 diabetes were more likely to receive antiplatelet agents (44%), ACE inhibitors (62%) than those without peripheral arterial disease (23%, p<0.001 and 49%, p<0.001 respectively)

(41)

33 but not statins (23% vs. 20%, p=0.12). These important cardioprotective medications are significantly underused in these populations.

1.11 Rationale

DECs are an important part of the diabetes care team (Jones et al., 2013) and will continue to be as the burden of diabetes increases in Ontario (Lipscombe et al., 2007).

DSME is designed to improve patient self-care in many areas including administration of medications to manage blood glucose, lipids and blood pressure, SMBG, and monitoring for complications (Ismail et al., 2004). There is evidence from RCTs that diabetes education improves glycemic control, blood glucose monitoring, blood pressure, weight and lipids (Ellis et al., 2004; Norris et al., 2001; Norris et al., 2002; Minet et al.,

2010). There is little evidence of the effect of diabetes education in a real world setting.

This study is designed to examine the effect of DEC attendance on prescription drug dispensation and retinopathy screening in routine clinical care in Ontario. Patients attending DECs learn about diabetes and its complications, and how best to treat it, and may become motivated to become advocates for their health. They may therefore press their primary care providers to improve quality of care.

By using drug prescription data for patients with diabetes it may be possible to draw an association between DEC attendance and quality of care. Large multi-center RCTs support the use of statins in most patients over 65 years with diabetes to decrease macrovascular risk (Collins et al., 2003; Colhoun et al., 2004). Statins were used as the primary outcome as in the over 65 population with diabetes most people would require pharmaceutical intervention to reach the target lipid levels making it a good measure of

(42)

34 quality of care. People attending DECs would learn about the importance of lipid

management in preventing diabetes complications and be more likely to accept or seek a prescription for a statin from their primary care provider. There is similar strong evidence to support the use of glucose lowering medications, antihypertensive medications and ACEI/ARBs in this population (UKPDS 33 and 34, Schrier et al., 2002; HOPE investigators 2000; Lindholm et al., 2002). DECs also review the importance of these interventions and they too can be used as indicators of quality of care and therefore they were included as secondary outcomes. As many patients are referred to a DEC to learn SMBG so this was also included as a secondary outcome. DEC attendance may serve as a reminder for the importance of regular retinopathy screening therefore it was also

included as a secondary outcome.

1.11.1 Primary Research Question

Do diabetic patients aged ≥ 65 years have increased filling of prescriptions for statins after attending a DEC in routine clinical care in Ontario?

1.11.2 Secondary research questions

Do diabetic patients aged ≥ 65 years have increased filling of prescriptions for glucose lowering medications, blood glucose monitoring, ACE inhibitor/ARB, antihypertensive medications, and retinopathy screening after attending a DEC in routine clinical care in Ontario?

1.11.3 Hypotheses

Diabetes education center attendance will be associated with increased use of statins in the over 65 population in Ontario. DEC attendance will also be associated with

(43)

35 increased the use of glucose lowering medications, antihypertensive medications, ACE inhibitor/ARB, SMBG, and retinopathy screening in this population.

DEC attendance will not be associated with increased use of proton pump inhibitors, levothyroxine and otolaryngology care (to test the specificity of the response from DEC attendance).

(44)

36

Chapter 2: Methods

2.1 Research Design

This is a population based cohort study of residents of Ontario aged 65 years or greater diagnosed with diabetes on or before January 1, 2005. It used health care administrative databases of the Ontario Ministry of Health and Long Term Care. These databases include the Ontario drug benefits program database (ODB) which contains the prescriptions filled under the provincial formulary for all residents aged ≥65 years; the physician service claims database of Ontario Health Insurance Plan (OHIP) which includes claims for fee-for-service reimbursement for all physician and optometry services provided in Ontario; the hospital discharge abstracts prepared by the Canadian Institute for Health Information (CIHI); and a demographic database (Registered Persons Database- RPDB) which includes birth and death dates, sex and postal code of home residence. Individual patients can be linked between all of these databases and across time via their reproducibly encrypted personal health card number.

DSME in Ontario is delivered mainly through 331 DECs throughout the province at academic or community hospitals, community health centers or First Nations

organizations and is funded in whole or in part by the Ministry of Health and Long Term Care. There was no previous administrative database or registry of DEC visits. Therefore, a registry of DEC attendance was created by collecting the visit dates and the health card numbers of all patients who attended any of the 331 DECs in Ontario in 2006 (Cauch- Dudek et al., 2013). Data were collected either manually from the DEC charts by trained

List of Abbreviations

ABSTRACT

ACKNOWLEDGEMENTS

Table of Contents

List of Figures and Tables

List of Abbreviations

Chapter 1: Introduction

Chapter 2: Methods